Models of metal toxicokinetics are critically evaluated using both newly generated data in the NoMiracle project as well as those originating from older studies. The analysis showed that the most frequently used one-compartment two-phase toxicokinetic model, with one assimilation and one elimination rate constant, does not describe correctly certain data sets pertaining particularly to the pattern of assimilation of trace elements. Using nickel toxicokinetics in carabid beetles and earthworms as examples, we showed that Ni in fact exhibits a three-phase kinetics with a short phase of fast metal accumulation immediately after exposure, followed by partial elimination to an equilibrium concentration at a later stage of a metal exposure phase, and by final elimination upon transfer to an uncontaminated food/soil. A similar phenomenon was also found for data on cadmium kinetics in ground beetles and copper kinetics in earthworms in data already published in the literature that was not accounted for in the earlier analysis of the data. The three-phase model suggests that the physiology of controlling body metal concentrations can change shortly after exposure, at least in some cases, by increasing the elimination rate and/or decreasing metal assimilation. Hence, the three-phase model, that allows for different assimilation and/or elimination rates in different phases of exposure to a toxicant, may provide insight into temporal changes in the physiology of metal handling. Consequently, this alternative model should always be tested when describing metal toxicokinetics when temporal patterns of internal metal concentration exhibit an initial "overshoot" in body metal concentrations. © 2009 Elsevier B.V.